Guide · Agent skill
How to use the ClickClack Ops agent skill (clickclack)
Updated · written with AI from the project's own documentation and checked against our catalogue data
The clickclack skill is a set of written instructions that teaches an AI agent how the ClickClack project is deployed and checked, so you can ask the agent to ship a build or confirm the live site is current. This guide explains what an agent skill is, how to install this one, how to trigger it with plain requests, and what it needs before it can do anything useful. It is written for maintainers and engineers on the ClickClack or OpenClaw team who work with this specific deploy setup.
Quick start
- 1Get the ZIP. Download the clickclack skill ZIP from this directory page and unzip it. Confirm the folder contains SKILL.md and agents/openai.yaml.
- 2Place the skill folder. Move the folder to ~/.claude/skills/ for a personal install, or to .claude/skills/ in a project. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Prepare the requirements. Make sure you have a local clone at ~/Projects/clickclack, pnpm, the Wrangler CLI, a Cloudflare login or API token, and Docker running. Access to the OpenClaw 1Password vault is also needed for the deploy token and R2 credentials.
- 4Ask for the task. Describe the task in plain language, such as deploying the latest build or checking a live domain. Be clear about whether you want a read-only check or a live deploy.
- 5Review the verification output. Check that the bundle hash on each live domain matches the newly deployed build and that the container health looks correct.
What is an agent skill and what does clickclack do?
An agent skill is a folder of instructions, usually a SKILL.md file, that an AI agent reads when a task matches the skill's description. It gives the agent project-specific knowledge it would not otherwise have.
The clickclack skill covers ClickClack product and runtime operations, Cloudflare Workers deploys, hosted app checks, and domain questions. ClickClack is a self-hostable Slack-style chat app. The skill tells the agent where the code lives, how to deploy it to Cloudflare Workers and a container, how to confirm the live site runs the new build, and which guardrails to follow.
- Knows the repo location and the public surfaces: the product site, the www and app domains, and the docs site.
- Walks through the deploy steps, including the container rollout gotcha.
- Lists verification checks for the live site, the container health, and the docs site.
- Enforces guardrails such as not printing OAuth secrets or magic tokens and not deploying from dirty git checkouts.
Install the clickclack skill from ZIP
Download the skill ZIP from this directory page and unzip it. The skill folder contains agents/openai.yaml and SKILL.md. Place the unzipped skill folder in one of these locations, depending on whether you want it for all projects or just one.
For a personal install available across projects, put it in ~/.claude/skills/. For a single project, put it in the project's .claude/skills/ folder.
In the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside the project folder
- Claude apps: Settings → Capabilities → Skills, then upload
mkdir -p ~/.claude/skillsTrigger the skill with natural requests
You do not need a special command. Ask for the task in plain language that matches the skill's description, such as deploys, hosted app checks, or domain questions. The agent should load the skill when the request fits.
Be specific about the outcome you want, for example whether you want a deploy, a verification, or only a read-only status check. The deploy steps change live infrastructure, so say clearly when you only want a check.
- Deploy the latest ClickClack main to Cloudflare and verify the live site.
- Check whether app.clickclack.chat is running the newest build.
- Confirm the docs site at docs.clickclack.chat is responding.
- Check the container health and image digest for the ClickClack container.
- Explain how the clickclack.chat and www domains are set up in DNS.
Deploy and verify a ClickClack build
The skill's deploy flow starts by updating the local clone, making sure Docker is running, and running the Cloudflare deploy script. Authentication can be a wrangler login through OAuth or the 1Password deploy token supplied as CLOUDFLARE_API_TOKEN.
After deploying, the skill has the agent compare the JavaScript bundle hash on the live site with the freshly deployed build. This confirms the public site is not still serving an old container instance.
- Update the clone with a fast-forward-only pull.
- Start Docker. The skill mentions OrbStack as one way to do this.
- Run the Cloudflare deploy script.
- Bump CLICKCLACK_CONTAINER_NAME in wrangler.jsonc so the running container is replaced, then redeploy.
- Commit the wrangler.jsonc bump.
- Check the bundle hash on each live domain and the container health with wrangler.
cd ~/Projects/clickclack && git pull --ff-onlyKnow the safety limits before you run it
The skill runs scripts and uses the network. Its deploy steps change live infrastructure that serves the public site, so a wrong token or an unclean checkout can affect real users. The skill is written for supervised use by people who know the project.
The deploy token has no DNS scope. The skill says DNS record changes must be made in the dashboard or with a token that has Zone.DNS permission. The skill also says not to print OAuth secrets or magic tokens, so review the agent's output if you are sharing a session.
- Do not use it for unattended or autonomous production deploys.
- Deploy only from clean git checkouts, not dirty working trees.
- Use the 1Password deploy token, which is limited to workers scope.
- Expect the agent to need network access to Cloudflare, the live domains, and GitHub.
What people use it for
- Shipping a new ClickClack build to production and confirming the public site is serving it.
- Checking whether app.clickclack.chat or www.clickclack.chat is running the latest bundle.
- Confirming the docs site at docs.clickclack.chat responds after a change.
- Reviewing the container health and image digest after a rollout.
Frequently asked questions
Is the clickclack skill free to use?
The skill is published under the MIT licence, according to the listing. You still need your own Cloudflare account access, the 1Password vault access, and the other tools listed as requirements.
What do I need to run the clickclack skill?
You need Git, a local clone of the ClickClack repository at ~/Projects/clickclack, pnpm and the Wrangler CLI, a Cloudflare login or API token, and Docker running locally. You also need access to the OpenClaw 1Password vault and network access to Cloudflare, the live domains, and GitHub.
Where do I put the skill so Claude can find it?
Put the unzipped folder in ~/.claude/skills/ for a personal install, or in a project's .claude/skills/ folder for one project. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Can I use this skill for my own ClickClack server?
Probably not as written. The skill is specific to one project and its infrastructure, and its paths, account IDs, and vault item names belong to that team. You would need to adapt the instructions to your own setup.
Why does my live site still show the old version after a deploy?
The skill explains that a running container instance is not replaced while it has traffic. Bumping CLICKCLACK_CONTAINER_NAME in wrangler.jsonc and redeploying forces a fresh instance on the new image. Then compare the bundle hash on the live site with the deployed build.
Can the skill change DNS records for me?
Not with the deploy token, which has no DNS scope. The skill says DNS record changes need the Cloudflare dashboard or a token with Zone.DNS permission.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you maintain ClickClack or work on its OpenClaw deploy setup and want an agent to follow the same runbook for deploys and live checks. Do not use it if you only want to use ClickClack as a chat app, if you are outside that deploy setup, or if you want unattended production deploys.
Download the skill